Hadoop-control is a project mainly written in Java, it's free.
A re-architecting of the Hadoop web applications open for all to contribute.
The purpose of hadoop-control (aka: another poorly named open source project) is to rearchitect the Hadoop core web applications around a modern MVC structure on top of Spring MVC. Ideally, this would, at some point, be submitted back to the ASF to be included in Hadoop proper.
Project hosting is currently at http://github.org/esammer/hadoop-control
This project is actively accepting any contributions and community support.
Build
Hadoop Control currently uses Maven as its build system.
# mvn package
Should produce a reasonable working package (by the end of 0.0.1) in the
form of a war file.Roadmap
0.0.1:
- Bootstrap a skeletal UI for the Job Tracker.
- Figure out how to, with minimal disruption, integrate with the
existing Hadoop code base. (Possibly defer actual integration until
the code reaches a reasonable point.)
- Create stubs for the JobTracker, Configuration, ClusterMetics, and
other core Hadoop objects to facilitate testing and prototyping.
- Build community support.FAQ
Q: Why use Spring? What about license issues?
A: Spring 3 is open source, licensed under APLv2, known by many Java devs,
is one of the nicer frameworks for web MVC apps and REST services.
Q: How does this fit in with the Hadoop core code base?
A: After some research into the way the web apps are bootstraped in Hadoop,
I found we can probably get away with building one or more wars and then
make small modifications to the existing Jetty code to load them rather
than the exploded JSP directories currently in the source. This also
breaks the web apps out of the code base proper further simplifying
builds and even allowing some folks to simply not deploy them at all.
Q: So are you aiming to replace all of the web UIs as well as the servlets
used to vend mapper outputs, etc.?
A: Only the UIs for now. The servlets and code vending the mapper outputs
and other RPC-ish HTTP stuff is far more "mission critical" and simpler
in business logic and benefits less from what this project is set out to
do.
Q: What set this off?
A: There's little (no?) input validation, proper handling of URL encoding
/ decoding, ability to inject authentication services, etc. It's
something to do. ;)Authors
E. Sammer <[email protected]>